Skip to main content

Full text of "Characterization and evaluation of a sealed cell mercury resonance ionization imaging detector"

See other formats


THE CHARACTERIZATION AND EVALUATION OF A SEALED CELL MERCURY 
RESONANCE IONIZATION IMAGING DETECTOR 












By 
MI HAEL RODNEY SHEPARD 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 

OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 

OF THE REQUIREMENTS FOR THE DEGREE OF 

DOCTOR OF PHILOSOPHY 

UNIVERSITY OF FLORIDA 

2002 



This dissertation is dedicated to the memory of my grandfather and best friend, 
Dean Clay Greer, and to the memory of my grandfather and mentor, John Shepard. 






ACKNOWLEDGMENTS 

In the years since I first came to the University of Florida I have overcome many of, 
perhaps, my life's greatest trials. More recently, I have been blessed with the arrival of a 
healthy and beautiful daughter, Hannah Grace Shepard. And now, as I approach the end 
of my graduate career, I can fully realize and appreciate all these personal and 
professional experiences. 

My success as a graduate student is to be attributed to all of those who have helped me 
along the way. With that said, my foremost thanks go to my research advisor, Dr. James 
Winefordner. His knowledge of the field and passion to teach provided a wonderful 
learning environment. His flexibility and sense of humor created a pleasurable work 
place. Finally, it was his sincerity and patience on those "rainy days" that made this 
degree attainable for me. 

Gratitude must also go to Dr. Benjamin Smith for his support. Ben's willingness to 
find the missing power supply or to diagnose an instrumental problem made him an 
essential resource. His positive attitude and encouraging words of wisdom make him an 
invaluable friend. 

Further gratitude is extended to Dr. Oleg Matveev for his numerous ideas and 
support of my research progress. Oleg helped to make sense of strange results and 
always knew of an alternative way make a certain measurement. 



m 



I would also like to thank Drs. Kathryn Williams, Robert Kennedy, Weihong Tan, 
Vaneica Young and Gardiner Myers for their support and direction during my teaching 
appointments. I am also grateful to the US Defense Advanced Research Projects 
Agency, the US National Institute of Health (Grant HL63965), the Graduate Student 
Counsel, and Dr. John Helling for funding support of my research and travel. 

My greatest support over the past five years has undoubtedly come from my wife, 
Michele. It is for that reason that I give my deepest thanks to her. Without her support 
and encouragement, I would not have found the motivation to complete this work. 

Finally, I would like to extend my gratitude to all the past and present JDW group 
members that I have worked with for their support and guidance. Special thanks go to 
those that I worked most closely with, including Jamshid Temirov and Drs. Ricardo 
Aucelio and Gabor Galbacs. 

Above all, I give thanks to God for all of the blessings in my life. 



IV 



TABLE OF CONTENTS 

page 

ACKNOWLEDGMENTS iii 

LIST OF TABLES vii 

LIST OF FIGURES viii 

ABSTRACT x 

CHAPTERS 

1 THE GOAL OF THE PROJECT 1 

2 INTRODUCTION TO RESONANCE IONIZATION IMAGING DETECTORS 3 

Background 3 

Historical Preface to Resonance Ionization Spectroscopy 3 

Ultra-narrowband Imaging Detectors 5 

Luminosity-Resolving Power Product 8 

Principles of Operation 9 

Modes of Operation 12 

Detection of Charged Particles 12 

Signal measurement 14 

Experimental Description 16 

Previous Resonance Ionization Imaging Detectors 20 

3 EVALUATION AND CHARACTERIZATION OF THE RIID 26 

Sources of Noise 26 

Spectral Bandwidth and Range 27 

Sensitivity 33 

Spatial Resolution 33 

Temporal Resolution and Response 34 

Quantum Efficiency 36 



4 TIME RESOLVED MEASUREMENTS IN THE RIID 38 

Photoelectric effect in the RIID 38 

Temporal Response of the RIID in Non-imaging Mode 39 

Effect of X2 and X3 Position within the RIID 39 

Effect of High Voltage Application 43 

Signal-to-Noise Ratio in Non-imaging Mode 45 

Unidentified Ionization Component 47 



5 IMAGE DISTORTIONS IN THE RIID 54 

Overview of Image Distortions 54 

Experimental 56 

Temporal Distortions 56 

Spatial Distortions 64 



6 POTENTIAL APPLICATIONS 67 

Moving Object Detection 67 

Monte Carlo Simulation of Moving Object Detection 69 



7 FINAL COMMENTS 83 

Conclusions 83 

Future Work 83 



LIST OF REFERENCES 85 

BIOGRAPHICAL SKETCH 90 



VI 



LIST OF TABLES 
Table page 

2-1 . Elements suitable for use in UBID systems 7 

3-1. List of elements suitable for use as the RIID's active medium 32 

6-1. MCS software variable definitions and default values 70 



vn 



LIST OF FIGURES 
Figure page 

2- 1 . Schematic representation of several atomic vapor ultra-narrowband imaging 

detectors 4 

2-2. Luminosity-resolving power product (adapted from Matveev et al. ) for several 

spectroscopic systems 10 

2-3. Two and three color ionization schemes for mercury 11 

2-4. Side view illustration of the RIID for imaging mode operation. X 2 and X 3 are 

perpendicular with X\ and the place of the paper 13 

2-5. Side view illustration of the RIID for non-imaging mode operation. X 2 and 1* are 
perpendicular with X\ and the place of the paper 15 

2-6. The sealed-cell mercury resonance ionization imaging detector 18 

2-7. High voltage divider for the RIID 19 

2-8. Experimental setup of the RIID 21 

2-9. Image quality versus high voltage across the voltage divider 22 

2-10. Imaging mask 23 

2-11. Prototype RIID with MCP 25 

3-1 . Spectral response of the RIID in non-imaging mode as A,i is detuned away from 

the center of resonance line 29 

3-2. Isotope and hyperfine splitting of mercury's ground state transition (6'S -> 

6 Pi) (adapted from Grossman et al.) 30 

3-3. RIID captured image with 80 urn spatial resolution 35 

4-1. Time resolved measurement in the RIID 40 

4-2 . RIID schematic showing \ 2 and A, 3 position and signal generation 41 



vin 



4-3. Non-imaging mode signal when the ionization region shifted away from input 

window toward the MCP 42 

4-4. Imaging mode detection for %i and X3 close to input window and MCP 44 

4-5. Effect of high voltage upon non-imaging mode signal 46 

4-6. Non-imaging mode signal-to-noise ratio of 135 for V = 9.0kV and 10 9 incident 

photons. ( — ) shows noise due to X2 and X3 beams only; ( — ) shows signal when 
all three beams are present 48 

4-7. Non-imaging mode signal-to-noise ratio of 64 for V = 9.5kV and 10 5 incident 

photons 49 

4-8. Peak separation as a function of Vmcp-iw 51 

4-9. Peak intensity ratio as a function of V M cp-iw 52 

5-1. Image series with V M cp-iw = 4.8kV 57 

5-2. Image series with Vmcp-iw = l.OkV 60 

5-3. Spatial Resolution as a function of time. • - V M cp-iw = 4.8 kV ; A - V M cp-iw = 

1.0 kV 61 

5-4. Signal conservation during image distortions 63 

5-5. Spatial image distortions in the RIID 65 

6-2. Schematic diagram of MCS model 71 

6-3. Simulation with V disk = m/s and v cen t e r =0 73 

6-4. Simulation with V dis k = 30.0m/s and v cent er =0 75 

6-5. Simulation with V disk = 30.0m/s and v cen ter =2580 xlO" 6 cm" 1 76 

6-6. Doppler shifted profiles 78 

6-7. Simulation with V disk - 30.0m/s, v cent er =2580 xlO" 6 cm" 1 , and u scat = 32.0 

mm" 79 

6-8. Simulation with V disk = 30.0m/s, v center =2580 xlO" 6 cm" 1 , u scat = u scat = 12.0 

mm" 1 81 






IX 



Abstract of Dissertation Presented to the Graduate School 

of the University of Florida in Partial Fulfillment of the 

Requirements for the Degree of Doctor of Philosophy 

THE CHARACTERIZATION AND EVALUATION OF A SEALED CELL MERCURY 
RESONANCE IONIZATION IMAGING DETECTOR 

By 

Michael Rodney Shepard 
August 2002 

Chair: James D. Winefordner 
Department: Chemistry 

In many fields of imaging science, there is an increasing demand for detectors with 

high sensitivity as well as high spatial, spectral, and temporal resolution. Although 

conventional imaging systems frequently excel in one of these aspects, it is often at the 

expense of the others. The mercury vapor filled Resonance Ionization Imaging Detector 

(RIID) presented here is not subject to this tradeoff. When operated in conjunction with 

modern narrowband lasers, the analytical figures of merit of the RIID can far exceed 

those of conventional imaging systems. Under certain operating conditions, the spectral 

resolution of the RIID can approach the natural atomic linewidth of the contained atomic 

vapor. This type of ultra-narrowband detector has the potential for a wide range of 

applications such as the detection of moving objects, the imaging of ultrasonic fields, 

plasma diagnostics, and high-energy particle detection, among others. Additionally, the 

RIID can be operated in a variety of modes providing multifaceted system information. 



A compact, sealed cell version of a mercury vapor RIID is demonstrated in this study. 
The figures of merit obtainable with this novel design are presented. The signal-to-noise 
ratio of the RIID in ion detection mode for both imaging and non-imaging cases was 
evaluated. Additionally, the temporal response of the RIID for both imaging and non- 
imaging modes was studied. 

A limitation of the compact mercury RIID is realized in the imaging mode. The 
operating conditions for optimal image signal-to-noise and spatial resolution in the RIID 
were shown to result in severe image distortions. Several methods for relieving such 
distortions are proposed. 



XI 



CHAPTER 1 
THE GOAL OF THE PROJECT 

In many fields of imaging science, there is an increasing demand for optical detectors 
with improved imaging figures of merit. The resonance ionization imaging detector 
(RIID) has potential applications in many of these fields and can offer improved spectral 
resolution and sensitivity with comparable spatial and temporal resolution. It has been 
suggested that the RIID could be used in a variety of applications ranging from ultrasonic 
field detection and chemical Raman Imaging to atmospheric projectile and satellite 
tracking [1]. The potential for this type of ultra-narrowband imaging detector has not yet 
been fully realized, nor has the technology been fully developed. 

The prototype RIID, experimentally demonstrated at the University of Florida in 1998, 
featured a low pressure mercury vapor imaging cell [2,3]. This prototype ionization 
imaging detector successfully demonstrated the sensitivity and selectivity of atomic 
vapor, ultra-narrowband detection principles. Additionally, it was shown that 2- 
dimensional (2-D) imaging information was obtainable with this type of detector [4]. 
However, the experimental design of this prototype RIID featured a series of vacuum 
pumps, a large atom reservoir, and an intensified charge coupled device (CCD) detector 
among a host of other bulky components. The colossal design of this prototype was not 
practical for the applications in imaging science as discussed above. In addition to these 
spatial constraints, other aspects of the prototype design limit the application of the 
detector. The cumbersome nature of turbomolecular vacuum pump system, the external 



atom reservoir, and the complex optical design of the system were all undesirable 
features of this initial design. 

The goal of this work was to characterize and evaluate the performance of a compact 
variant of the mercury resonance ionization imaging detector. A cell was designed and 
manufactured to contain all of the necessary components of the RIID into a sealed and 
compact detector. The performance of this RIID was evaluated using two separate 
ionization schemes for comparison. Both schemes were based upon the ground state 
absorption of 254 nm photons. The behavior of the RIID for both imaging and non- 
imaging modes of detection was investigated. The RIID was also studied for both ion 
and electron detection modes. 






CHAPTER 2 
INTRODUCTION TO RESONANCE IONIZATION IMAGING DETECTORS 

Increasing demands in imaging science have brought forth the development of 
spectrally selective, atomic vapor imaging detectors and filters [5-7]. The photon 
detector evaluated in this work unites the well-established principles of resonance 
ionization spectroscopy with modern imaging science and related fields. 

Background 

A new generation of ultra-narrowband imaging detectors (UBIDs) can offer spectral 
resolutions limited only by the natural atomic linewidth of the contained atomic vapor. 
Compared to conventional imaging techniques, the use of UBIDs can result in a 
resolution improvement of 2-4 orders of magnitude [1,8,9]. Figure 2-1 illustrates three 
types of UBIDs and the respective instrument function of each. Additionally, UBIDs 
demonstrate the ability to detect low signal levels in the presence of a high background. 
One such UBID, the resonance ionization imaging detector (RIID), can provide a spectral 
resolution on the order of 1 .4 MHz when an atomic mercury vapor is used as the medium 
[10]. 

Applications which may require this type of high spectral resolution and sensitivity 
include laser Doppler velocimetry, ultrasonic field imaging, and moving object detection. 
Potential applications of the RIID will be discussed later in this dissertation. 
Historical Preface to Resonance Ionization Spectroscopy 

Resonance ionization spectroscopy (RIS) was first proposed in the early 1970s by 
Letokhov for isotopic speciation and trace metal detection [11]. The first experimental 



Resonance ionization imaging 

10-20 kV 
y 



Instrument function 



i 



m 



CCD Camera 




Atomic vapor cell 



Av=10kHz-30GHz 



Resonance fluorescence imaging 

Broadband color filters 



'4 



t 



CCD Camera 




1FL 



Av=10 kHz - 30 GHz 



tkj Atomic vapor cell 



Imaging with an atomic absorption Filter 

Atomic vapor cell 




Av-10kHz-30GHz 



Magneto-optical imaging 

Polarizers 







Atomic vapor cell 
m magnetic field 



CCD Camera 




Av=2 - 30 GHz 



Figure 2-1 . Schematic representation of several atomic vapor ultra-narrowband imaging 
detectors. 



demonstration of RIS was performed in 1971 at the Institute of Spectroscopy of the 
USSR Academy of Sciences [11-13]. This experiment featured a 2 step photoionization 
scheme for rubidium atoms. A tunable dye laser, pumped by a Ruby laser, was used to 
first excite the ground state Rb atoms into the 5 2 ?>/ 2 state. The second-harmonic output 
pulses of the Ruby laser were then sufficient to photoionize the excited state Rb atoms. 
The selectivity of this scheme was demonstrated as the energy of the second-harmonic 
output pulses of the Ruby laser were insufficient to photoionize the ground state Rb 
atoms. The authors cited an overall ionization efficiency of about 0.1%. 

By 1977, the success of this newfound ionization technique brought forth the first 
analytical experiments for detection of single atoms [11]. Hurst et al, at Oak Ridge 
National Laboratory, used a two-step photoionization scheme for the detection of single 
cesium atoms in a buffer gas [11,14-15]. 

Since these first experiments, resonance ionization schemes and supporting 
experiments have been reported for atoms of nearly every element [11,15-18]. The 
underlying principles of RIS have been applied to molecular methods for improved 
selectivity as well [11,14]. Most notable is the coupling of RIS with mass spectrometry 
for techniques such as resonance enhanced multiphoton ionization mass spectrometry 
(REMPI-MS) and resonance ionizing mass spectrometry (RIMS) [19-22]. The principles 
of RIS are applied in this work, not for the optical detection of Hg atoms but for the 
detection of photons by a Hg vapor. 

Ultra-narrowband Imaging Detectors 

Atomic vapors have very narrow absorption lines making them ideal candidates for 
use as narrowband optical filters or as active media in UBIDs. When monoisotopic 
atomic vapors are used, or when Doppler-free techniques are employed, the spectral 



response of the detector or filter is further improved. In the case of a mercury vapor 
UBID, the spectral resolution can be improved from over 1.0 GHz to 1.45 MHz. The 
spectral resolution of conventional imaging systems is at best 20-50 GHz [1]. Only 
photons, whose frequency falls within this exceptionally narrow linewidth, will be 
absorbed by the contained atomic vapor for eventual detection. The inherent selectivity 
and spectral resolution of UBIDs arise from this fact. Selectivity is further added to this 
mode of photon detection when additional transitions are utilized for detection as is the 
case with the RIID and resonance fluorescence imaging monochromator (RFIM). Such 
atomic vapor imaging detectors have limited spectral ranges corresponding to the 
absorption frequencies of a few volatile elements. The availability of laser systems 
further limits the application of this technology. At present, there are at least 23 elements 
that are suitable for use in most UBID systems including Cs, Hg, Rb, and Sr [1,5,11]. 
These elements are shown in table 2-1. In the case of the mercury RIID, the high 
sensitivity results from the final ionization step and by the amplification factor of the 
microchannel plate (MCP) [23]. 

In 1996, Matveev et al. demonstrated single photo-electron and photon detection using 
a non-imaging mercury resonance ionization detector (RID) [23]. The RID cell was 
developed to detect low photon levels following the avalanche ionization of a buffer gas 
contained within the mercury cell. A limit of detection of 253.7 nm photons, 
corresponding to the 6'S -> 6 3 Pi transition, was shown to be 0.5 quanta during the 
lifetime of the excited state. 

A naturally occurring mercury vapor was chosen in this work to be the active medium 
of the RIID for several reasons. The vapor pressure of mercury provides a saturated 



7 



Element 


Minimal Cell 
Temperature (°C) 


Hg 


-59 


Cs 





Rb 


17 


K 


42 


Cd 


94 


Na 


99 


Zn 


148 


Mg 


207 


Yb 


231 


Li 


271 


Sr 


275 


Ca 


319 


Eu 


231 


Tl 


335 


Ba 


353 


Pb 


383 


Sm 


413 


Tm 


491 


He 


N/A 


Ne 


N/A 


At 


N/A 


Kr 


N/A 


Xe 


N/A 



Table 2-1. Elements suitable for use in UBID systems. 



8 

atomic cell at room temperature (4 X 10 13 atoms-cm" 3 ) . The concentration of gaseous 
mercury atoms would be sufficient to absorb more than 90% of the resonant photons 
within an optical path length of 2-3 cm [24]. Additionally, the ultraviolet and visible 
wavelength transitions are easily obtainable with current lasers systems in our 
laboratories. 

Luminosity-Resolving Power Product 
Conventional imaging systems frequently excel in one aspect of image acquisition, 
while suffering at others [25,26]. The RIID, and UBIDs in general, are not subject to this 
tradeoff. A figure of merit which best represents the performance of spectroscopic 
imaging systems is the luminosity-resolving power (LR) product, otherwise known as the 
spectral efficiency of the imaging system. Resolving powers (R) on the order of 10 6 are 
achievable for almost all conventional spectroscopic imaging devices. However, the 
throughput of the device, or luminosity (L) (cm 2 -sr), is inherently decreased in order to 
achieve these high resolving powers. For this reason, the product of the luminosity and 
resolving power (LR) provides more information about a system than does each figure of 
merit reported separately. The relationship between LR and the signal-to noise ratio 
(S/N) is also beneficial when evaluating spectroscopic imaging systems. Matveev et al. 
derived an equation relating these figures of merit [25]. This relationship is given in 
equation 2-1. 



Pe 



S/N = ^J-rk— /**(*&«« (2-d 

\B„fiA p J 



9 

So is the maximum value of the spectral-detection function, B^b (photons s "ST "Ottk ) is 
the background level, p is a optimal resolution proportionality constant, A (nm) is the 
mean wavelength of incident photons, P is the number of pixels, s is the ratio of the 
image detector's working area to the total area,/? is the effective number of pixels (P » 
p), Bxs(A.) (photons I'-arim") is source spectral radiance, and S n (^) (nm) is the 
normalized spectral-detection function. The LR product for several imaging systems has 
been evaluated by Matveev et al. [26]. Figure 2-2 compares the LR product for many 
conventional imaging systems with the RIID. As shown, the RED has a much greater LR 
product compared to popular imaging systems. 

Principles of Operation 

The principles of operation for the resonance ionization imaging detector studied here 
are based upon the resonant photoionization of a isotopic mixture of mercury atoms. In 
this work, three-step ionization schemes (two or three color) were used where the final 
transition was non-resonant photoionization. The two schemes employed are shown in 
figure 2-3. The analytical beam, or probe beam, to be imaged was the first UV transition, 
denoted \\. This beam is directed through the input (front) window (IW). The absorption 
of A.] (253. 7nm) is a resonant transition from the 6'S ground state to the 6 3 Pi excited 
state. From the 6 3 Pi level, a second resonant photon (k 2 ) at 313.2 nm (6 3 Pi -> 6 3 Di) or 
435.8 nm (6 3 Pi -> 7 3 Si) was introduced through the side window. From this point, 
photoionization was achieved by a third photon (k 3 = 626.4 nm, 6 3 Di -> Hg + ; X 3 = 435.8 
nm, 7 Si — » Hg + ) non-resonant transition via the side window. 

Upon ionization, the electron/ion pair experience a high electric field (externally 
applied potential) and are accelerated toward opposite ends of the cell. The polarity of 



10 



10' 



10 u 






3 10* 





io- 



10 



10 



Michel* 



Infrared heterodyne detection LOurm 



UV heterodyne detection 0.25 (am 



1 1 i — i i i 1 1 1 1 1 



~i — i — i i i 1 1 1 1 i — i i i i 1 1 1 1 1 — i — r-r 



10' 



10 B 



10' 
Resolving power 



10" 



10 9 



Figure 2-2. Luminosity-resolving power product (adapted from Matveev et al. ) for 
several spectroscopic systems [25]. 



11 



(10.57 9V) (10 l 3eV) 

k\\\\\\\\ SSSSSSSS^SSSSSS ^^^ l (1 0.44 e V) 



435.8nm (*J 



626.4nm (X,) 



7^ (7.731 eV) 



435.8nm (^) 




6^ (8.345 eV) 



.2nm p^) 



6^ (4.887 eV) 



Figure 2-3. Two and three color ionization schemes for mercury. 



12 

the electric field accelerates the positively charged ions toward the MCP for eventual 
detection. Before detection, however, the ion signal is amplified and converted by the 
MCP, into an electronic signal. These electrons, now representing the analytical signal, 
are accelerated in a second electric field onto a luminescent screen. The resulting 
luminescence is then detected by a conventional CCD camera. A schematic drawing, 
representing a side view of the RIID, is shown in figure 2-4. 

For the applications of the RIID, X\ can be directed into the RIID for immediate 
detection, transmitted through an imaging mask into the RIID, or reflected off of an 
object toward the detector. In the latter two cases, X\ will take the shape of the imaging 
mask or the shape of the scattering or reflecting object. 

Modes of Operation 

The resonance ionization imaging detector investigated here can be operated in a 
variety of "modes," each providing unique information about the input signal or 
ionization events within the actual detector. 
Detection of Charged Particles 

Upon the ionization of mercury atoms in the RIID, a high voltage (V M cp— iw) is 
applied between the MCP and front input window, or simply the atom cell. The resulting 
electric field accelerates the charged pair to opposite sides of the cell. The relative 
polarity of this applied field determines the directionality of each component. The 
mercury ions migrate toward the more negative region. Conversely, the photoelectrons 
move toward the more positive region. When the MCP is held at a more negative 
potential than the input window, ions are accelerated toward the MCP for ultimate 
detection and, hence, the RIID is said to be in ion detection mode or normal mode. 



13 



Input Window 



>■■=> 




Thin Metal Film 



MCP 



Luminescent Screen 



eO 
e O 



O 



y 



i 






Side Window fc, A,) 



Figure 2-4. Side view illustration of the RIDD for imaging mode operation. \% and fa are 
perpendicular with "k\ and the place of the paper. 



14 

The counterpart of this technique, in which the photoelectrons are accelerated toward the 
MCP, is appropriately termed electron detection mode or reverse mode. Although there 
is no clear analytical advantage of electron detection mode, it was crucial to 
understanding some of the limitations of sealed-cell design RIID. This will be discussed 
in greater detail in the chapter 5. 
Signal measurement 

Perhaps of greater interest are the two forms of signal transduction in the RIID, 
namely imaging and non-imaging modes. Unlike the above distinction between particle 
detection modes, both methods of signal transduction provide useful information about 
signal generation. The data obtained from each detection mode are unique and can be 
collected simultaneously. 

Imaging mode, as described above and shown in figure 2-4, relies on electronic to 
optical signal conversion. Electrons generated at the MCP are accelerated onto a P-20 
type phosphor screen, common to conventional image intensifiers. The green 
luminescence from this screen can be seen visually or collected with a conventional CCD 
camera for further analysis. This mode allows 2 and 3-dimensional imaging information 
to be obtained about the incident 253.7 nm photons and their spatial distribution. 

Non-imaging mode bypasses the necessity of electronic to optical signal conversion. 
To minimize surface charging, a thin film of metallic platinum was coated on the inner 
side of the phosphor screen. Instead of grounding this metal film, as originally intended, 
an oscilloscope was used to measure the current as electrons from the MCP pass through 
the film onto the phosphor screen. Therefore, the RIID is only slightly altered for non- 
imaging mode measurements, as shown in figure 2-5. An obvious drawback to this 
operational mode is the loss of 2-dimensional imaging information. This drawback is 



15 



Thin Metal Film 



Input Window 



■Mi 



• Hg 
OHg + 
e- e- 






e "°rSZ 



o~ 



T 




Luminescent Screen 



Side Window {X 2 , 3L) 



Oscilloscope 



Figure 2-5. Side view illustration of the RIID for non-imaging mode operation. X 2 and 
X3 are perpendicular with X\ and the place of the paper. 



16 

easily negated as the RIID can be simultaneously operated in both imaging and non- 
imaging modes. The usefulness of this mode will be discussed later in chapter 4. 

Experimental Description 

The design of this sealed-cell mercury RIID is the product of several prototype 
systems to be discussed in the upcoming section. The RIID in this work was constructed, 
according to our design, by NPP Radian (Moscow, Russia). The compact cell is 5 cm in 
diameter and 4 cm in width, with a front input widow diameter of 2.5 cm and side 
window widths of 0.7 cm. In this study, X2 and X^ are focused with a cylindrical lens 
into a rectangular sheet approximately 0.1 mm in width. X\, which carries the imaging 
information in these experiments, is expanded to a 1.5 cm 2 spot size. Images were 
created by placing the imaging mask between the X\ beam expander and the front input 
window of the RIID. Amplification of the ion signal was achieved with a chevron-type 
microchannel plate of approximately 10 5 -10 6 gain. The actual amplification factor of this 
MCP is unknown. The P-20 type ((Zn,Cd)S:Ag composition) phosphor screen used was 
approximately 10 urn thick and was capable of quantum efficiencies up to 50%. To 
prevent surface charging of the input window and phosphor screen due to constant 
electron bombardment, thin metal films (5-10 nm) were coated onto the inner surfaces of 
each. The input window was coated with a palladium film and the phosphor screen was 
coated with a platinum film. The films were also used to maintain homogenous electric 
fields between each component and the MCP. Mercury was introduced to the cell, in 
excess, in the form of a titanium-mercury alloy dispenser, or Getter type dispenser [27]. 
All of the components listed here were sealed under a slight vacuum (< 1 atmosphere). 
The input window and phosphor screen are shown in figure 2-6. It should be noted that 



17 

there are several uncertainties about the construction of this RIID. These included the 
actual MCP amplification factor, the true thickness of the phosphor screen and metal 
films, the composition of the mercury dispenser, the exact pressure within the cell, and 
the quantum efficiency of the phosphor screen. A single high voltage power supply, 
coupled to a voltage divider, provided the necessary voltages to the stages of RIID 
detection. Figure 2-7 illustrates the high voltage divider. 

The three-step photoionization schemes shown in Figure 2-3 were achieved with two 
dye lasers pumped by a XeCl Excimer laser with a pulse energy of 100 mJ at 308 nm and 
10 Hz operation (Lambda Physik, model LPX-240i, Acton, MA). Xj at 253.7 nm (6'So 
— » 6 3 Pi ) was generated by frequency doubling Coumarin 500 laser dye (Exciton, 
Dayton, OH) in a Molectron dye laser (Molectron, Portland, OR). The remaining steps 
for the two color ionization scheme, %$ =435.8 nm (6 3 Pi -* 7 3 Si) and X} = 435.8 nm 
(7 3 Si — » Hg + ), were generated by Coumarin 120 laser dye (Lambda Physik, Acton, MA). 
For the three color ionization scheme, X 2 =313.2 nm (6 3 Pi -> 6 3 Di) and X3 = 626.4 nm 
(6 Di -> Hg + ), Rhodamine 101 laser dye was used (Lambda Physik, Acton, MA). When 
working with the two color ionization scheme, pulse energies of 10 uJ and 450 uJ were 
typically measured for X\ and X 2 ( and X 3 ), respectively. Pulse energies of 10 uJ, 7.2 uJ, 
and 140 uJ were typically measured for X\, X2, and A.3, respectively, for the three color 
ionization scheme. The dye laser used for X 2 and A,3 generation was constructed from a 
Lambda Physik Scanmate dye laser (Lambda Physik, Acton, MA). The laser was 
constructed as a mode-free, broad band dye laser. The result was an efficient dye laser 
which was continuously tunable over 70 nm and had a linewidth of 14 cm' 1 . 



18 




Luminescent screen 



^^■■■■■■■■HHHH^^^^^^^^HIMHHi^HHi 



Figure 2-6. The sealed-cell mercury resonance ionization imaging detector. 



19 






Input Window MicroChannel Plate 




Phosphor Screen 



vV\A 




vAA/^ AAA- 

35 MQ 86 MQ 



Figure 2-7. High voltage divider for the RIID. 



20 

The experimental setup described here is shown schematically in Figure 2-8. Time 
resolved measurements were made with a Tektronix oscilloscope (Tektronix model Tek 
TDS3012, Beaverton, OR). 

Shown in figure 2-9 are a typical series of images obtained as a function of high 
voltage. As shown in this series of images, there exists an operating voltage for optimal 
signal-to-noise and spatial resolution. The optimal voltage in figure 2-9 appears to be 
between 9.0 and 9.5kV. The imaging mask employed in throughout this study is shown 
in figure 2-10. Images were collected in real time with a CCD camera and transferred to 
a desktop PC via a National Instruments image acquisition board (model EVIAQ, PCI- 
141 1, Austin, TX). The monochrome CCD camera featured 510(H) x 492(W) pixels and 
a 0.04 Lux sensitivity (Supercircuits, model PC23C, Leander, TX). 

Previous Resonance Ionization Imaging Detectors 

As mentioned earlier, the RIID is still in the earliest stages of development. The first 
demonstration of the RIID was in 1998 [2]. The detector design in those first 
experiments had several drawbacks including image distortions and noise limitations due 
to scattered radiation within the cell. The absence of a microchannel plate also limited 
the sensitivity of the detector. Nevertheless, the principle of RIID operation was proven. 
The authors were able to demonstrate that the low pressure version of the mercury RIID 
was capable of acquiring images from two different wavelengths, namely 253.7 nm and 
435.8 nm. 

Shortly after this proof of principle, the RIID was improved by the addition of a 
microchannel plate and shorter cell length [4]. With these improvements, the first 
observation of 2-dimensional image detection with an atomic vapor imaging detector was 
made. The spatial resolution obtained with this detector was on the order of 0.2 mm. 



21 





■ 


- 






Figure 2-8. Experimental setup of the RIID. 






22 




Figure 2-9. Image quality versus high voltage across the voltage divider. 






23 




5.0 mm 



Figure 2-10. Imaging mask. 






24 

Surface charging and, thus, image distortions were the primary limitations. The detector 
described here is shown in figure 2-11. When comparing figures 2-6 and 2-11, the 
motivation behind this project becomes clear. 

Following further improvement, a distortion free mercury RIID was demonstrated 
[28]. Spatial resolution on the order of 120 um was achieved. These two 
accomplishments were the result of coating the inner surface of the input window with a 
thin Pt film. The 10 nm thick film reduced transmission of X\ by 30%, but improved 
imaging characteristics on several fronts. The film acted as a contact between 
accumulated surface charge on the input window with ground. Additionally, a more 
homogenous electric field was created between the input window and microchannel plate. 
This greatly improved the spatial resolution of the detector by forcing the ions to migrate 
in unaltered paths toward the MCP. The problems overcome with this version of the 
mercury RIID will be revisited in chapter 5. 



25 




Figure 2-11. Prototype RIK) with MCP. 



CHAPTER 3 
EVALUATION AND CHARACTERIZATION OF THE RIID 

The resonance ionization imaging detector described here provides both sensitive and 
selective photon detection. The figures of merit for the mercury RIID make it a 
comparable, and in many cases superior, imaging technique for a wide range of 
applications. This chapter will describe the figures of merit for the RIID developed in 
this study and RIIDs in general. Most of the figures of merit for the RIID described here 
can, of course, be modeled after those of the resonance ionization detector (RID) 
[1,23,29]. 
Sources of Noise 

One possible source of noise within the RIID is the non-selective photoionization of 
atoms and molecular dimers [1,11,29]. In the case of a mercury filled RIID, molecular 
dimers do not form and can be omitted from this discussion [11,21]. As shown in figure 
2-3, the final step for mercury ionization is a non-resonant transition. In this case, 435.8 
nm photons are used for the photoionization from the 7 3 Si excited state. It is possible, 
however, for photons with wavelengths shorter than 458 nm to also photoionize the 
excited state mercury atoms. As a result, non-selective ionization could occur and its 
associated signal would be detected. Although this source of noise was not observed for 
the RIID presented here, its potential certainly exists. Various optical filters were used to 
show that the ionization only occurred as a result of A, 3 = 435.8 nm. 

Another source of noise in the RIID is the generation of electrons, via the 
photoelectric effect. In this case, X\ strikes a metal surface of a given work function and 

26 



27 

an electron is ejected. These electrons may be ejected from the MCP or from the metal 
film coating on the input window [2]. When the detector is operated in ion detection 
mode, only the electrons generated at the MCP are problematic. This effect will be 
discussed in detail in the following chapter. 

A final source of noise in imaging mode RIID operation is background luminescence. 
This background luminescence is more pronounced at high voltages and is observed as 
random "spots" on the phosphor screen. Because this luminescence is observed even 
without incident laser radiation, this noise likely arises within the final stage of detection 
[31]. A probable explanation of this noise is autoelectronic emission from the MCP [31- 
33]. By operating at lower voltage, when X\ is sufficiently high, this source of noise can 
be eliminated [34]. 
Spectral Bandwidth and Range 

The most significant advantage of the RIID, when compared to conventional imaging 
detectors, is the selectivity with which photons are detected. In the case of the RIID, the 
figure of merit which best represents selectivity is the spectral bandwidth (s) [35]. The 
spectral bandwidth of the RIID is limited by the absorption linewidth of the contained 
atomic vapor. For that reason, it is useful to consider the spectral bandwidth as a measure 
of the background rejection of the detector. Photons which are not within this spectral 
bandwidth are rejected by the detector. 

For the mercury RIID described here, the Doppler broadened linewidth of the mercury 
vapor is about 25 GHz. This linewidth is achieved by summating the linewidths of 
mercury's 7 stable isotopes. It has been calculated that the background rejection for such 
an atomic vapor, at a frequency shift of 0.6 cm" 1 , should be on the order of 10" 3 % [1,29]. 



28 

The observed frequency response, corresponding to the spectral bandwidth, is shown in 
figure 3-1 when the RIID is operated in the non-imaging mode. The level of background 
rejection, when X\ is detuned by 0.6 cm" 1 , is about 45 %. This level improves to over 97 
% at a 4.0 cm" 1 frequency shift. When operated in imaging mode, these figures of merit 
are not as impressive. At a frequency shift of 0.6 cm" 1 , the rejection level is 30 % and 
only about 65 % at 4.0 cm" 1 . This is most likely due to photoelectric signal generation, 
which is not influenced by small frequency shifts as shown here. The photoelectric and 
resonance ionization signal cannot be discerned in the imaging mode. A discussion of 
photoelectric signal generation will be presented in the following chapter. 

The observed 25 GHz linewidth for the isotopic mixture, used in this work, is the 
result of Doppler broadening and hyperfine level splitting. Figure 3-2 illustrates the 
isotopic and hyperfine splitting components of the 6'So -> 6 3 Pi transition at 253.7 nm 
[36]. The natural atomic linewidth of the mercury isotopes shown in figure 3-2 is on the 
order of 1.4 MHz. However, Doppler broadening can increase each to about 1 GHz. 
Doppler broadening of atomic lines results from the statistical distribution of velocities of 
the absorbing atoms [35]. The Doppler effect causes a distribution in the measured 
frequencies that is directly related to this velocity distribution. When Doppler-free 
techniques are employed with a monoisotopic vapor cell, the spectral bandwidth can 
approach the natural atomic linewidth of a single mercury isotope of 1 .4 MHz. Doppler- 
free techniques refer to experiments in which the velocity, and thus frequency, 
distribution is minimized in some fashion [37]. Such techniques include saturation 







29 


50^ 






■ 






40- 


i 





30 

E 

ro 20 

c 

m 

10- 



0- 







i 
3 



i i 



4 



■« r 

5 



Frequency Shift (cm' 1 ) 



Figure 3-1. Spectral response of the RIID in non-imaging mode as X\ is detuned away 
from the center of resonance line. 






CM 

o 

CM 



30 



25 GHz 



CM 






COO O) 

mun cm 
coco CO 






Av(10"W) 



8 



o 

I 



CD 



O 
CM 

J_ 



00 o 


t^ 


CO 


o 


CO CO 


O) 


in 


as 


T— T- 


CM 


CO 


p5 



Figure 3-2. Isotope and hyperfine splitting of mercury's ground state transition (6'S 
6 3 Pi) (adapted from Grossman et al.) [36]. 



-> 



31 

spectroscopy, lambda dip spectroscopy, and collimated molecular beam spectroscopy 
[35,38]. However, the RIID design discussed here does not permit Doppler-free 
techniques to be applied. Future RIID designs could be designed to allow such Doppler- 
free measurements. 

When Doppler-free measurements are not accessible, differential imaging techniques 
can also be used to improve upon the spectral bandwidth. Differential imaging, as 
applied to the RIID, refers to a technique in which image intensities are measured as a 
function of X\ frequency. Image intensity will increase as X\ approaches the center of the 
resonance transition. For example, it has been show that 80 MHz frequency resolution is 
achievable with UBID differential imaging techniques [39]. In that study, a fluorescence 
scheme of photon detection was used rather than ionization. 

The spectral working range of a particular RIID is inherently small, being limited to 
the linewidth of the first atomic transition. For the mercury RIID, the maximum working 
range is 25 GHz at 253.7 nm, which corresponds to 253.7 ± 0.0002 nra. This narrow 
band is responsible for the selectivity of the RIID, but limits the working range for many 
applications. By using a different atomic vapor, each with unique X] and s, the working 
range of the RIID could be shifted and, perhaps, slightly improved. When both laser 
availability and engineering limitations are considered, there are 10 elements suitable for 
use as the RIID's active medium [1,11,12]. These elements are listed in table 3-1. 






32 



Element 


Ionization Energy (eV) 


X-i (nm) 


Reference 


Li 


5.35 


670.8 


[40,41] 


Na 


5.14 


589.0, 589.6 


[42-44] 


K 


4.34 


404.7, 766.5 


[45,46] 


Ca 


6.11 


422.7, 616.2 


[47,48] 


Rb 


4.18 


420.2, 780.0, 794.8 


[49-51] 


Sr 


5.69 


460.7, 689.3 


[48] 


Cs 


3.89 


459.3,852.1,894.3 


[52] 


Ba 


5.21 


553.5,791.1 


[53-55] 


Hg 


10.44 


253.7,312.8 


[56,57] 


Tl 


6.11 


276.8, 377.6 


[48] 



Table 3-1 . List of elements suitable for use as the RIID's active medium. 



33 

Sensitivity 

The high sensitivity of most RID and RIID systems arises from the final, non-resonant 
ionization step. The efficiency for this photoionization step can approach 100 % 
[11,22,58]. By saturating the 6 3 Pi excited state (see figure 2-3), each 253.7 run photon 
absorbed should then be detected. Using a three-step resonance excitation scheme, 
followed by collisional ionization in a buffer gas, Matveev et al. achieved single photon 
detection in a RID [23]. It is therefore possible that single photon detection limits might 
be achieved with the RIID. 

Although the Hg RIID presented here did not achieve this lower detection limit, 
relatively low light conditions were detected. When operated in the imaging mode, fewer 
than 1,000 incident photons were detected via image summation [31]. When A-i intensity 
is low, it was necessary to sum several images (10 - 20) to improve the S/B. In the case 
of 16 image summations, the S/B is improved 15 times. 

When operated in non-imaging mode, a lower photon detection limit can be achieved. 
A 5.0 mV signal can be discerned from the noise, which corresponded to a S/N of 3. 
This, in turn, correlated to about 540 incident photons. In this case, the X.i beam was 
focused into a small absorbing volume for more efficient ionization. When X\ was 
expanded for 2-D experiments, the limit of detection of incident photons was degraded to 
about 900. These types of experiments will be covered more thoroughly in the following 
chapter. 
Spatial Resolution 

The spatial resolution obtainable with conventional CCD cameras and image 
intensifies is on the order of about 30 urn [33,59]. It has been calculated that the RIID 



34 

could achieve comparable results [1,10]. This is realized as the second stage of RIID 
detection (MCP to phosphor screen) is identical to that of conventional image intensifiers 
[60]. For various reasons, ranging from cell engineering to imaging distortions, this 
degree of spatial resolution has not been achieved. However, for the RIID described in 
this work, a spatial resolution of 80 um has been observed. Shown in figure 3-3 is an 
image obtained with this detector, corresponding to a spatial resolution of 80 um. 

Several improvements to cell design could be made to improve upon the obtainable 
spatial resolution. Such include a shorter flight path for ions to the MCP and a shorter 
distance between the MCP and phosphor screen. In principle, the phosphor could be 
coated directly onto the MCP. However, as will be shown in upcoming chapters, there 
are disadvantages of employing these types of improvements. 
Temporal Resolution and Response 

In principle, the time resolution of the RIID (ACrqd) is limited by two factors: the 
flight time of the ion from generation to the MCP (x F ) and the rate of ionization (s" 1 ) due 
to the additional laser radiation (Radd)- 1 Equation 3-1 relates these terms 
mathematically. 



^■^RIID ~ ■\j T F + 



' 1 ^ 



V ^ADD J 



(3-1) 



When ionization occurs close to the MCP, a voltage can be applied such that x F can be as 
low as 200 ps. From this, it is apparent that the ultimate limit upon temporal resolution in 
the RIID will be the rate of ionization. The rate of ionization by the second and third step 





35 






1.3 mm 






Figure 3-3. RIID captured image with 80 urn spatial resolution. 



36 

photons is limited by the lifetime of their respected excited state. For the ionization 
scheme shown if figure 2-3, the 7 3 Si state has a lifetime of 8 ns [61]. For this reason, the 
Hg RIID described here could achieve a temporal resolution of 8 ns. However, as in the 
case of spatial resolution, experimental limitations prevent this observation. The 
temporal resolution for the detector described in this study was limited to the frequency at 
which the ionizing lasers were operated (1-10 Hz). 
Quantum Efficiency 

The following discussion of quantum efficiency will pertain to RIIDs in general, rather 
than the detector described in this dissertation. Experimental limitations stemming from 
the design of the sealed cell detector prevented a direct measurement of this figure of 
merit. 

Quantum efficiency (q) for the mercury RIID is defined as the ratio between the 
number of ions created to the number of input Xj photons. A more practical definition is 
shown in equation (3-2). 

q = arj (3-2) 

The quantum efficiency is the product of a, the absorption factor of the atomic vapor, and 
?], the ionization efficiency; rj is defined as the ratio of the number of ions created to the 
number of 7 Si excited state mercury atoms. Values of 0.1 have previously been 
obtained for a mercury filled RID [23]. In this study, the authors used relatively weak 
lasers for this measurement. When higher energy lasers are employed (> 1 mJ/pulse), 



37 

there are no fundamental limitations preventing an ionization efficiency near 100 % 
[1,12,36]. 

The maximum value of a will, of course, occur at the center of the absorption line. 
The absorption factor a is defined by the following equation. 

a = \-e- nai (3-3) 

In equation 3-3, n is the density (cm* 3 ) of the atomic vapor at a given temperature, cris 
the cross section (cm 2 ) for the ground state transition, and / is the optical path length 
(cm). In the case of the Hg RIID at room temperature, n = 4 X 10 13 atoms-cm" 3 and a = 6 
X 10" 13 cm 2 . 

Under these conditions, it has been calculated that a quantum efficiency greater than 
90 % can be achieved for a mercury filled RIID [1]. It has also been shown that a 
quantum efficiency greater that 60 % can be achieved for at least 23 elements below 300 
°C [12] (see table 2-1). As discussed above, atomic vapors other than mercury would 
broadened the working range of the detector. The possibility for near unity quantum 
efficiency further illustrates the capabilities of the RIID as a sensitive photon detector. 



CHAPTER 4 
TIME RESOLVED MEASUREMENTS IN THE RIID 

Non-imaging mode operation of the RIID involves a current measurement rather than 
an image capture. Upon the resonant photoionization of the contained mercury vapor, the 
signal ions are accelerated toward the MCP for amplification and eventual detection. 
This electronic signal is converted into an optical signal, via the phosphor screen, or 
detected directly. The electronic signal is measured as electrons from the MCP pass 
through a thin metal film in route to the phosphor screen. This platinum thin film, 
normally grounded to relieve surface charging, in this case, is connected to an 
oscilloscope. Because these electrons still reach the phosphor screen, both non-imaging 
and imaging modes can be simultaneously performed. 

Photoelectric effect in the RIID 

A key advantage of the RIID is its ability to selectively detect minimal photons in the 
presence of high background. However, a source of noise in the RIID is the signal 
generation due to the photoelectric effect (PE) by high levels of X\ signal photons. In 
experiments with an intense X\ source, not all 253.7 nm photons are absorbed by the Hg 
vapor. In this case, X\ is transmitted onto the surface of the MCP and the photoelectric 
effect is observed. The MCP response to these PE electrons is identical to that of the 
mercury ions. An image can be created on the phosphor screen, which corresponds to 
this second source of signal (from transmitted X\). This is shown experimentally when X 2 
and A.3 are blocked from the RIID. Without these final transitions, mercury ionization 



38 



39 

will not occur. The resulting image is created entirely by electrons generated from the 
photoelectric effect on the surface of the MCP. Under normal operating conditions, when 
the three beams (X\, X 2 , A.3) needed for ionization are present, the image formed contains 
both PE and ionization components. These two components of the signal cannot be 
discerned in the imaging mode. 

Temporal Response of the RIID in Non-imaging Mode 

The non-imaging mode is useful to resolve the PE and resonance ionization 
components of the signal. Shown in figure 4-1 is a time resolved measurement, typical of 
non-imaging mode RIID detection. The location within cell, where each signal 
component is generated, is given by the measured time. This measurement can be 
thought of as a "flight" time for the ion to reach the MCP. Since the PE signal is 
generated on the surface of the MCP, t = is assigned to the PE peak. Ionization of the 
mercury vapor takes place at some depth within cell, so the ionization component will 
have a longer travel time. For the example shown in figure 4-1, Hg ionization occurred 
in the center of the RIID. From that point, it took the mercury ion 350 ns to reach the 
MCP. The peak detected at approximately 100 ns will be discussed later in this chapter. 
Effect of X 2 and X 3 Position within the RIID 

Figure 4-2 illustrates how the location of the ionization region can be varied with the 
position of X 2 and X.3. The flight time of the Hg ion can, of course, change as the sheet of 
X 2 and A, 3 is shifted between the input window and MCP. Shown in figure 4-3 is the non- 
imaging RIID signal when the ionization region is shifted away from the input window 
toward the MCP. A difference of nearly 75 ns in flight times is observed for the 
resonance ionization signals in this case. 



40 



0- 



-10- 



-20 



V 

% -30 



-40- 



-50- 




\ 



Unidentified 



Photoelectric signal 



Resonance Ionization signal 




100 200 300 

Time (ns) 



400 



500 



Figure 4-1 . Time resolved measurement in the RIID. 



41 






Input window 



Ionization 
Photoelectric 



MCP 




Phosphor screen 



A^ & X i position 






Figure 4-2. RIID schematic showing Xj and X 3 position and signal generation. 












42 



0.000- 



-0.002- 



0> 

a 

> -0.004- 



-0.006- 




lonization Region 

Close to Input Window 

Center 

Close to MCP 



T 




— I - 
200 



Time (ns) 



400 



Figure 4-3. Non-imaging mode signal when the ionization region shifted away from input 
window toward the MCP. 



43 

There are several reasons for varying the position of the ionization region in this 
fashion. One such reason, as discussed above, is to allow the separation of PE and 
ionization signal. This is achieved by having the ionization region close to the input 
window, such that the flight time of the mercury ion is relatively long. Another reason to 
have the ionization region close to the input window is improved ionization efficiency, 
and thus a lower photon limit of detection. This arises from the fact that most X\ photons 
are absorbed within the first few millimeters of the absorption cell. X2 and X3 should also 
be directed into this region of the cell for most efficient ionization. 

The primary advantages of having the ionization region closer to the MCP are 
observed in the imaging mode of detection. The spatial resolution of the RIID can be 
improved when ionization events occur close to the MCP. Although steps have been 
taken for a homogenous electric field between the input window and MCP, even slight 
differences in this accelerating field will result in image distortions. Decreasing the flight 
path of the ion will decrease the probability of ion diffusion within the cell and, hence, 
improve spatial resolution. Figure 4-4 shows images in when the ionization region is 
moved between the input window and MCP. A similar problem to this one, namely 
image distortions, will be also be minimized when ionization occurs close to the MCP. 
These phenomena will be discussed in detail in the following chapter. 
Effect of High Voltage Application 

During the first stage of detection in the RIID, a high voltage (V M cp-iw) is applied 
between the input window and microchannel plate. The electric field created between 
these two components acts as an acceleration field for the formed ions. 



44 





Input Window 



MicroChannel Plate 






Figure 4-4. Imaging mode detection for X 2 and X3 close to input window and MCP. 



45 

The flight time, in seconds, of the ion is inversely proportional to the electric field 
strength [9]. 



-ylr/zY'Av c 4 - 1 ) 



This relationship is shown in equation 4-1, where W is the mass to charge ratio (kg) of 

the mercury ion, L is the length of the flight tube (or width of the RIID) (m), and V is the 
applied voltage (volts). It is predicted from this equation that the flight time of the 
mercury ions will be reduced as V M cp-iw is increased. This is shown experimentally in 
figure 4-5. The ionization region is held constant in the center of the cell in this example, 
while Vmcp-iw is varied. Similar to the example shown in figure 4-3, the flight times of 
the ion signals are altered. As the V M cp-iw is increased between 0.95 and 1 .40 kV, the 
flight times are varied between 624 and 682 ns. These are much longer flight times as 
compared to the example in figure 4-3, in which V M cp-iw = 4.5 kV. The magnitudes of 
both the PE and ion peaks are also altered. This was is expected when the voltage 
scheme shown in figure 2-6 is used, as the gain of the MCP is increased as V M cp-iw is 
increased. When the RIID is operated in the imaging mode, this may result in a limiting 
source of noise. 

Signal-to-Noise Ratio in Non-imaging Mode 

An advantage of non-imaging mode detection is an improved S/N. This is achieved 
by the temporal resolution of the ionization signal from the PE background. Upon 
resolution of these signal components, the limiting noise is reduced by several orders of 
magnitude. A discussion of limiting noise sources was described in the previous chapter. 



46 



0.000- 



-0.001 - 



O) -0.002- 

I 



-0.003 - 



-0.004 








MCP-IW 



1.40 kV 
1.15 kV 
0.95 kV 



— i ' 1 - 

250 500 

Time (ns) 



- 1 - 

750 



Figure 4-5. Effect of high voltage upon non-imaging mode signal. 



■Q : 



47 

A given number of incident photons can be achieved by attenuating X\ with neutral 
density filter combinations. Figure 4-6 shows the detection of approximately 10 9 incident 
253.7 nm photons. In this case, a S/N of 135 is observed. Shown in this figure is non- 
imaging signal as well as the electronic noise measurement when X\ is not present. Noise 
is calculated as the standard deviation of the latter case. When the number of incident X\ 
photons is decreased to 10 5 , the applied voltage must be increased from 9.0 kV to 9.5 kV. 
As shown in figure 4-7, the S/N is degraded from 135 to 64. It is implied from these data 
that the applied high voltage ultimately limits the S/N and, thus, the limit of photon 
detection in the non-imaging mode. 

Unidentified Ionization Component 

As shown in the previous figures of this chapter, an additional signal component exists 
between the PE and ionization peaks. Time was devoted in this work to the identification 
of this peak. However, experimental limitations and uncertainties concerning the cell's 
construction have not allowed a positive identification. Furthermore, this unidentified 
component is only discernable in the non-imaging mode. Possibilities of the origin of 
this peak are deduced from non-imaging mode data shown here. 

The hypothesis for these experiments arises from flight time calculations with 
equation 4-1. In this hypothesis, it is assumed that the signal of interest originates within 
the atom cell of the RIID. Secondly, the unknown ion causing this signal must have a 
lower m/z than does the mercury ion. Possibilities for a lower m/z include singly charged 
ions, such as sodium and potassium ions, or doubly charged mercury ions. 

The possibility of doubly charged mercury ions is not likely, based only on a few 
experiments. As shown above, the flight times for ions to reach the MCP can be varied 



48 




T 
150 300 

Time (ns) 



450 



Figure 4-6. Non-imaging mode signal-to-noise ratio of 135 for V = 9.0kV and 10 9 

incident photons. ( — ) shows noise due to X2 and A.3 beams only; ( — ) shows signal when 
all three beams are present. 



49 



-0.006 H 




150 
Time (ns) 



450 



Figure 4-7. Non-imaging mode signal-to-noise ratio of 64 for V = 9.5kV and 10 5 incident 
photons. 






50 

with high voltage. Figure 4-5 illustrates the effect of high voltage upon flight times. 
From equation 4-1 it is predicted that Hg + and Hg +2 will have different flight times, but 
should change proportionally with each other as the applied voltage is varied. In other 
words, the separation between these two peaks should remain constant as the high voltage 
is increased. The relative intensity of each peak should also change proportionally as 
MCP gain is increase. However, this behavior is not experimentally observed. Figure 4- 
8 shows the separation of these two peaks as a function of Vmcp-iw- From this plot, we 
see that the flight time of the resonance ionization signal is influenced more by voltage 
than is the unknown component. In figure 4-9, the intensity ratios of these peaks is 
shown as a function of voltage. Similar to the previous example, the intensity of the 
resonance ionization signal is affected more by voltages changes than is the unknown 
component. These two experiments show that unknown signal component is not likely 
due to doubly charged mercury ions. 

Several important implications can be extracted from these data. The unknown signal 
component originates from a positively charged ion. Proof of this is provided by the 
experimental observation of detecting this component only in the ion detection mode. 
Any negatively charged species would be accelerated toward the input window. 
Furthermore, as proof against a doubly charged mercury ion suggests, this positive ion 
must have a lower m/z than mercury. Calculations show that the molecular weight of 
these ions were in the range of 20-55 amu assuming singly charged ions. This wide 
range of molecular weights does not allow further deductions. One final consideration is 
the response of the unknown signal component to X u X 2 and X 3 photons. It was observed 
that this peak was dependent upon both A.] and X 2 (and X 3 ). In the absence of X u as 



51 



tov - 














700- 


i 












650- 














-600- 

c 550- 
O 

2 500- 

re 

a 

5 

v> 450- 

re 

i. 400- 




i 


























i 


i 






350- 










i 
















i 


300- 




1 ' 


i ■ 






1 i ■ 



200 



400 



600 



800 



1000 



1200 



MCP-IW 



(V) 



Figure 4-8. Peak separation as a function of V M cp-iw- 



52 






5.0 -| 
4.5- 
4.0- 



1 3.5 ^ 



w 3.0 

c 

9 

1 2.5 J 



1 



2.0- 
1.5- 



1.0 












400 500 600 700 800 900 

V (V) 

MCP-IW v ' 



1 ' 1 ' 1 

1000 1100 1200 



Figure 4-9. Peak intensity ratio as a function of V M cp-iw- 



53 






expected, no signal component is detected. A different observation is made when X] is 
present, but detuned away from the center of the absorption line. The PE signal remains 
constant, but the ionization signal components are reduced as k\ is tuned further away 
from the transition. Similar observations are made when X2 is eliminated or detuned. 
These data show that the unknown signal component is not formed via multiphoton 
ionization by the 435.8 run photons of Xj or related processes. The unidentified signal is 
generated by some process involving 253.7 nm photons. 

In summary, the identity of the ion responsible for the third non-imaging mode signal 
component remains unknown. The ion responsible is most likely positively charged and 
of low molecular weight. The ion is probably the result of some impurity introduced 
during cell construction. Possible impurities include sodium and potassium. Cell 
limitations did not allow the spectroscopic validation of these impurities. 



CHAPTER 5 
IMAGE DISTORTIONS IN THE RIID 

The motivation behind the development of the RIID is the demand for a detector with 
the highest possible spectral resolution and 2-dimensional imaging capabilities. The 
current RIID design meets this demand and is capable of a spatial resolution of < 80 urn. 
As might be expected with any technology in the earliest stages of development, the RIID 
does have its share of limitations. The primary limitation for imaging mode RIID is 
image distortions [62]. Spatial resolution for a given experiment can be degraded over 2 
orders of magnitude. An extensive investigation of these distortions and their origins 
provide insight toward the development of future RIID cells. 

Overview of Image Distortions 

In typical RIID experiments, an accelerating voltage is applied between the input 
window and microchannel plate, such that the positively charged mercury ions are 
accelerated toward the MCP for eventual detection. As a result of the polarity of the 
electric field, the electrons produced during ionization are accelerated toward the input 
window. After continuous periods of operation, charge may accumulate on the surface of 
the input window. The result of such surface charging is an altered electric field between 
the MCP and input window. The pathways followed by the ions traveling to the MCP are 
redirected and, in turn, the spatial distribution of the signal is altered. The redistribution 
of signal ions is observed experimentally as image distortions. 

An improvement to the design of the RIID detectors, which eliminated this type of 
imaging artifact, has been previously reported [63]. This design enhancement involved 

54 



55 

coating the inner surface of the input window with a thin palladium film. Although this 
metal film decreased the transmission of X\ through the input window by about 35%, 
image distortions were practically eliminated. 

There are, however, observed limits to this improvement technique when implemented 
into the compact cell design discussed in this dissertation. Firstly, the non-uniformity of 
the metal film may allow different degrees of surface charging, or more likely, different 
rates of charge removal. Such film heterogeneity may allow variations in the electric 
field between the MCP and input window resulting in image distortions. Variations in 
metal film thickness may result from poor deposition methods or from highly energetic 
electron bombardment of the metal surface. Secondly, when high accelerating voltages 
are used, high energy electrons may pass through the metal film and penetrate into the 
quartz input window. The penetration depth of a high energy electron into a pure 
material of known density can be calculated by the empirical equation 5-1 [64,65]. 



D = °— (5-1) 



The penetration depth D is given in urn when E (kV) is the accelerating voltage 
experienced by the electron and p (g-cm" 3 ) is the density of the material. This expression 
assumes a 5.0 mm distance between the stationary electron and surface. For a 10 nm 
thick Pd film under the given constraints for equation 5-1, an electron could penetrate the 
input window to a depth of 0.5 urn when V M cp-iw = 5kV. Penetration of an electron into 
the quartz input window would result in the same type of electric field distortion, 
resulting in image distortions. 



56 

Experimental 

The two color ionization scheme shown in figure 2-3 was employed in this study. X\ 

- 253.7 nm (6'S -» 6 3 Pi) and X 2 = X 3 - 435.8 nm (6 3 P, -> 7 3 S, -» Hg + ) with 
measured pulse energies of 10 uJ and 450 uJ, respectively, were used. It was shown that 
the behavior of the RIID, in terms of distortion effects, was unaffected when the three 
color ionization scheme in figure 2-3 was used. 1 A sheet of X 2 and X 3 was directed into 
the side window of the RIID approximately 0.25 cm from the input window. 

Temporal Distortions 

The intensity of the resonance ionization imaging signal under normal operating 
conditions should depend upon two experimental variables only. It is directly 
proportional to the pulse energies of the lasers and the magnitude of the applied voltages. 
However, it has been observed that under constant experimental conditions, image 
quality and intensity can vary with continuous operation. This is most readily observed 
when high V M cp-iw are applied. Typically, a voltage of 4.5-5.0 kV is applied between the 
MCP and input window for an optimal signal-to-noise ratio and spatial resolution. 
Nonetheless, these figures of merit are significantly impaired by image distortions with 
long operating times. 

Figure 5-1 shows a typical series of images obtained as a function of time with V M cp- 
iw = 4.8 kV. The initial image (t=0) remains unchanged for approximately 6 minutes at 
which point image degradation is readily observed. For most applications, signal or 
image measurements would be made within this 6 minute window. For the 



57 



.. ■ ..> \ '•■ -■..■ . v ■ ; ■-■-,!■ ";■ :, 




t = 



Hik 1 



r 



t = 5 minutes 



:■:.,■■ ■■ ■ 

T 




t= 10 minutes 



t= 15 minutes 





t = 20 minutes 



t = 25 minutes 



Figure 5-1. Image series with V M cp-iw = 4.8kV. 



58 

detection of minimal X\ photons, when signal integration or image averaging is likely, 
this time frame may not be sufficient. 

Cycling the power supply on and off upon the recognition of image artifacts does not 
recover the initial image. In fact, this distortion effect has been observed for several 
hours after the power supply has been shut off. It is implied from this data that some type 
of charging effect is involved. The charge will dissipate in most cases after 4-5 hours, at 
which point the image distortions are no longer observed. 

Further proof of input window surface charging is realized when a "neutralization" 
potential is applied. By reversing the polarity of the electric field between the MCP and 
input window (i.e., electron detection mode which involves making the MCP more 
positive than the input window), image distortions are relieved within seconds rather than 
hours. Experimentally, this was accomplished by removing the input window from 
electric ground and connecting it within the 120 MQ resistor chain shown in figure 2-6. 
There are two mechanisms by which the surface charging is relieved in this case. The 
first mechanism involves the acceleration of the accumulated electrons away from the 
input window (toward to the MCP). A second possible mechanism is the neutralization 
of the negative charge by positively charged mercury ions as they are then accelerated 
toward the input window. A combination of these mechanisms is most likely. The 
duration of this reversed field is on the order of 1-2 seconds. Experimental limitations 
prevent an accurate measurement of the reversed field duration. Experimental limitations 
also prevent signal measurements, imaging or non-imaging, during this period. 

As discussed above, the purpose of the metal film coating on the input window, when 
connected to electrical ground, was to minimize the accumulation of charge. A possible 



59 

limitation of the metal film is realized when high energy electrons pass through the film 
and are accumulated between the film and window. When the kinetic energy of the 
electron is great enough, penetration into the window could occur. This possibility was 
evaluated by collecting images with a lower V M cp-iw- As is shown in figure 5-2, image 
distortions are not observed when a lower Vmcp-iw (1.0 kV), than in the previous 
example, is applied. Equation 5-1 above supports these experiments. From equation 5-1, 
it was calculated that an electron would not have sufficient energy to travel through the 
thin metal film. A minimum Vmcp-iw for electron penetration through the metal film was 
estimated to be 1.64 kV. These data imply that the distortions shown in figure 5-1 are a 
result of high energy electrons passing through the metal film, thus voiding its benefit. 

It should be noted that the S/N and spatial resolution of the initial image (t=0) are 
degraded here compared to the previous example. However, they do remain constant 
through the duration of the experiment. Decreased S/N and spatial resolution at lower, 
non-optimized voltages as expected. The spatial resolution obtained for both low and 
high voltage experiments is shown graphically in figure 5-3. It is shown here that spatial 
resolution is not compromised by temporal distortions for the low voltage experiments. 
The spatial resolution is not optimal, but does remain constant for the duration of the 
experiment. The distortion effect for the higher voltage experiments, as a function of 
time, is linear. A distortion rate constant (r) of 38 unvmin" 1 is obtained for the high 
voltage experiments. This constant was obtained from the slope of the of the curve 



60 



- 




i 







t=0 



t = 5 minutes 








t = 1 minutes 



t= 15 minutes 








t = 20 minutes 



t = 25 minutes 



Figure 5-2. Image series with V MC p-iw = l.OkV. 



61 



800 n 



700- 



-=* 600- 
O 500 H 



° 400 H 

0) 

DC 

1 300 4 

OJ 

a 

CO 

200-1 



100 H J 







I 



5 




r = 38(jm/min 



i ■ i ■ i 
10 15 20 

Time (minutes) 



25 



Figure 5-3. Spatial Resolution as a function of time. • - Vmcp-iw = 4.8 kV ; A - V M cp-iw 
-l.OkV. 



62 

shown in figure 5-3. As discussed above, and as shown in figure 5-3, that r = for up to 
10 minutes of continuous operation. 

Input X\ photons are distributed in the shape of an imaged object or, in the this work, 
as the shape of an imaging mask. Under ideal conditions, ionization will occur in a like 
arrangement and eventual signal detection will follow as such. When the electric field 
between the MCP and input window is altered, this ideal situation is not observed. The 
image distortions shown in figure 5-1 are the result of the signal "redistribution" in the 
atom cell. Although altered pathways occur between the ionization region and the MCP, 
each ion generated in the atom cell is still detected. The proof of this signal conservation 
during image distortions can be illustrated when the RIID is simultaneously operated in 
both imaging and non-imaging modes. Figure 5-4 shows that the magnitude of the 
resonance ionization signal remains constant during distortion conditions. Less that 5 % 
difference was measured for non-imaging mode signals measured at 2 and 20 minutes. 

In summary, it is shown that conditions for optimal image S/N and spatial resolution 
in the RIID may result in severe image distortions. These distortions occur when the 
RIID is operated at high V M cp-iw for extended periods of time. 

Relief of imaging distortions is possible by allowing the charge to naturally dissipate 
with time, which can require up to 6 several hours. Almost instant distortion relief is 
possible with the application of a neutralization potential (V M cp-iw). Furthermore, 
distortions may be prevented completely by operating the RIID at lowered voltages. 
However, the elimination of distortions in this fashion are at the expense of image S/N 
and spatial resolution. 



63 



200- 



400- 



2 600- 



800- 



1000- 



V =4.8kV 

MCP-IW 



i-02 minutes 
t= 10 minutes 
t = 20 minutes 



T 





- 1 • 

200 

Time (nS) 



Resonance Peak 



400 



Figure 5-4. Signal conservation during image distortions. 






64 

Other methods to minimize, or even prevent, these imaging distortions are possible. 
One such method would be to employ lower repetition rate lasers and, thus, less frequent 
image acquisitions. In such a fashion, the ion/electron pair would be created less 
frequently and the rate of charge accumulation would be reduced or even prevented. A 
second method might incorporate electronics to alternate the system between ion and 
electron detection modes. Since image distortions occur after about 6 minutes of 
continuous operation, the mode switching frequency should be on the order of about 3 
mHz. In this case, the accumulated charge would be "neutralized" before image 
distortions occurred. 

Spatial Distortions 

Image distortions can be also be observed with high levels of incident X\ photons and 

low voltages, when the position of X 2 and Xt, in the side window is varied. This type of 
imaging artifact results only from X 2 and X3 position and not the extent of the operational 
period. Figure 4-2 illustrates the variable position of X 2 and X 3 in the side window of the 
RIID. When the sheet of X2 and X3 light is brought relatively close to the input window, 
within 2.0 mm, short term distortions occur. In other words, X 2 and X3 can be moved 
back into their original positions without prolonged image distortions. Figure 5-5 shows 
such distortions as a result of the position of X 2 and X3. The results here are consistent 
with those discussed in the case of temporal distortions. By moving the ionization region 
close to the input window, a dense region of electrons is created close to the input 
window. Consequently, the electric field between the input window and the MCP is 
altered and image distortions are observed. However, unlike the temporal 



65 








D = 2.5mm 



D = 2.0mm 








D= 1.5mm 



D = 1.0mm 





D = 0.5mm 



Contact 



Figure 5-5. Spatial image distortions in the RIID. 



66 

distortions discussed above, this case is reversible. Simply moving the ionization region 
away from the input window minimizes these imaging distortions. 



CHAPTER 6 
POTENTIAL APPLICATIONS 

There are numerous potential applications that exist for the resonance ionization 
imaging detector because of its high spectral resolution and sensitivity. These 
applications are most numerous throughout the fields of analytical science, such as 
plasma diagnostics, ion mobility spectrometry, Raman spectrometry, and particle size 
distribution measurements for combustion analysis. Other possibilities include 
combustion product imaging, aerodynamic field flow imaging, ultrasonic field imaging, 
as well as a host of biomedical, military, and atmospheric applications. Of particular 
interest is the application of moving object detection. Almost any type of moving object 
could, in theory, be studied with the RIID: combustion products in flames, aerodynamic 
flow fields, and even macroscopic projectiles. Figure 6-1 depicts a general experimental 
setup that might be used for such applications. 
Moving Object Detection 

The premise behind moving object detection is that X u detuned from the resonance 
line, will illuminate a moving object and be scattered back to the RIID. Detection of the 
moving object will occur when the scattered illuminating frequency is Doppler shifted 
back into resonance with the mercury. A similar application might involve a molecule 
shifting X] back into resonance with mercury via a Raman shift. Although conventional 
imaging systems are capable of detecting the rather large shifts associated with the 
Raman effect, the high sensitivity of the RIID could improve Raman imaging techniques. 



67 



68 



T3 

- 



& 



<u 



~r 



<^ 



2. 



S3 

o 

09 






- = 


* a 


« ac 


— o 


^ f) 


5 


1? 





"-I 



I 



if 



8 




> 

H 


U 








u 


. 




P4 


■ 




Figure 6-1. Experimental setup for RIID applications. 



69 

Monte Carlo Simulation of Moving Object Detection 

Jelalian has previously described the detection of moving objects upon the earth's 
surface with laser radar techniques [66]. This was achieved when the background 
scattering radiation was practically eliminated using coherent laser radar, which itself is 
considered a scanning imaging technique [67]. Comparatively, the RIID could detect 
these moving objects with greater speed and with much more valuable target recognition 
information without scanning. The following summary of this Monte Carlo simulation 
(MCS) study will show that the UBID should be an extremely sensitive detector for a 
variety of moving objects in turbid media. 

The Monte Carlo program used in this study was developed at St. Petersburg 
University, Russia. The model, upon which the program is designed, consists of four 
planes and is defined as the following: the plane of the moving disk, the plane of the 
turbid media surface, the plane of the lens, and the RIID plane, as depicted in figure 6-2. 
The distance between the lens and the medium surface is equal to the distance between 
the lens and the RIID. Scattering is assumed to be isotropic in the plane perpendicular to 
photon propagation and distributed in accordance with Henyey-Greenstein relationships 
[68]. When a photon is scattered from the moving disc, a Doppler shift occurs. The 
variable parameters of this simulation software are defined in table 6-1. All calculations 
are performed in the photon's system of coordinates. In the beginning of every 
elementary action, a random path length (L R ) for each photon generated. This length is 
then compared with l/p. a bs and 1/fXscat • The photon is considered to be "alive", or still 
propagating, if L R < l/u. a bs. This same photon, of path length L R , can undergo scattering 



70 






Variable 


Default value 


Definition 


C^inc 


90.0° 


Angle of incidence of laser beam with respect 
to the plane of the medium surface. 


Vdisc 


30 m-sec' 1 


Disk velocity 


^center 


2580xl0" 6 cm"' 


The spectroscopic shift of the illuminating 
laser radiation. 


M-scat 


0.50 mm" 1 


Scattering coefficient; scattering centers per 
unit length. 


Rdisc 


10 mm 


Disk radius. 


^photons 


10 b 


Initial number of photons. 


Focus 


30 mm 


Focal length of lens. 


K av er 


10 cm" 1 


Absorption coefficient of the RIID. 


"•laser 


760 nm 


Wavelength of incident radiation. 


ga 


0.5 


Anisotropy factor. 


Iriid 


- 


Random path length in RIID. 


L 


1.0 mm 


Depth of disk in turbid medium. 


Lr 


- 


Random path length in turbid medium. 


Uabs 


0.50 mm" 1 


Absorption coefficient; absorbing centers per 
unit length. 


5v 


100.00 xlO" 6 cm" 1 


Spectroscopic absorption range of the RIID. 


t-'laser 


30.0 mm 


Diameter of the laser beam. 


J^acts 


9000 


Number of scattering events per photon. 


Radius 


20.0 mm 


Radius of lens. 


RIID W jdih 


1.0 cm 


Width of the RIID. 



Table 6-1. MCS software variable definitions and default values. 



71 



h 



RIID width 



: g£ f J Sggs . ,^.>.^Vi W < g J, 
1 -V a ■!■■: v ;= =■■ i Lfia& fc S $ l -V 5 • : ■■: a 



Lens 



Turbid media surface'- 



1 1 ■ . . 1 i ■ . , i i ' ' < 



Laser radiation 




Rdisk 



inc 



Disk 

velocity 

vector 



Figure 6-2. Schematic diagram of MCS model. 



72 

events if Lr> l/(J. sca t. If Lr< 2L sin(aj nc ), then the photon is randomly scattered from the 
disc and away from the medium. At this point the photon can be detected by the RIID if, 
of course, the angle at which it is leaving is within the field of view of the lens. This is 
shown graphically in figure 6-2. Angle coordinates of this photon are then recalculated 
to obtain the final coordinates of the photon upon the RIID surface. The resulting image 
is that of the disk. The photons which remain active, but within the turbid medium, can 
again be scattered or absorbed. In principle, a photon can collide with the disc and 
scattering particles several times. In this case, the linewidth of the scattered radiation will 
be much greater than laser spectral linewidth. For every photon that reaches the RIID, 
the random path length (Irhd) is generated and then compared with the existing RIID 
width. If 1 R hd < RIID w idth a photon is considered to be detected. 

For simplicity, these simulations were performed for short distances between the 
moving object and RIID and for a laser wavelength of 760 nm. If desired, these 
simulations could be applied to long distances as well. However, there will be more 
photons which "die" or are scattered away from the detector over a larger distance. To 
compensate for this, the number of initially emitted photons should be increased. A 
value of 9000 for N acts was selected after the realization that higher values do not render 
any noticeable changes. 

The first simulation, shown in figure 6-3, shows the MCS image of a stationary object 
in a turbid medium with small scattering and absorption coefficients. No distinct features 
are reveled from this image for two reasons. In this first simulation, the MCS parameter 
for v center is zero. This implies that a if a Doppler-shifted frequency from any 



73 




Focu* 
* avet 
J.|a«t 


|30.0 


|10.0 


(7800 



Figure 6-3. Simulation with Vdisk = m/s and Vccter =0 






74 

moving object was present, namely the disk or medium, it would not have been detected. 
Secondly, and more obvious, the disk velocity in this example is zero. Regardless of the 
Vcemer parameter value, a stationary disk would not produce a Doppler-shifted frequency. 
In summary, a scattering object is detected, but cannot be resolved. This example might 
represent the image of a stationary object by a conventional detection system. 

The parameters in simulation 2 are identical to those in the previous simulation, except 
that the disk velocity is 30.0 m/s rather than 0. Again, there is no recognizable object in 
the MCS image shown in figure 6-4. However, one significant change is observed from 
the first simulation. The number of photons detected is greatly reduced. Also, there is a 
vacancy of signal from the center of the image in the form of a disk. Such a result is 
considered a "negative" image. This image arises from the Doppler shift of the moving 
disk and v cen ter = 0. 

When the shift parameter v cen ter is changed to the predicted value of the Doppler shift, 
as shown in simulation 3, complete "resolution" of the moving disk is observed (figure 6- 
5). Resolution in this context refers to the unmistakable shape of the image and its 
observed diameter, and should not be confused with resolution as an analytical figure of 
merit. The spectroscopic shift parameter, v cen ter , is easily calculated from the Doppler 
equations for a disk of known velocity. 

V cen,er = <^0A I Kser ) «*&* I 2 ) (6" 1 ) 

As one might anticipate from the direct proportionality of this equation, a larger disk 
velocity results in a relatively large spectroscopic shift from the center of the incident 



75 



E 
E 



su 










40 
30 
20 


o 


•■ m 

Q u O 




o 
P o 


10 








JS ° 
*5d * 


U 
10 


9 M 


StsEB&Sa^i 




R °° 


20 
30 


°c 

• 


WJgP 


PS5 6 


• 


40 








. - _ -\ 


sn 


• 









-50 



50 









mm 






% Parameter s 






degree 
mm/sec 
mm ' 

10 G cm " 1 
nun 

mm 

™ 1 

cm 
nm 


?a 

L 

Cabs 
& 

Diaset 
Nacts 
Ratfers 

width 




mm 
mm 

10' G cm " 1 
mm 

mm 
cm 


ttinc 


90.0 


0.5 


v disk 


30000.0 


1.0 


Neat 


0.50 


0.10 


^center 


0.00 


100.00 


Rdsk 


10.0 


30.0 


N photons 


100000 


9000 


Focus 


30.0 


20.0 


kave, 


10.0 


1.0 


^lasef 


780.0 



Figure 6-4. Simulation with V<fek = 30.0m/s and Venter =0 






76 























50 
40 
30 
20 
10 

i ° 

-10 
-20 
-30 
-40 
-50 







i 
i 






— , 

— i 

— i 






• 








1 










50 






mm 






50 




\ Parameters 






degree 

mm/sec 

mm 

10" 6 cm " 1 

mm 

mm 
cm 
nm 


L 

Cabs 

b> 

Dlaset 

Nads 

Radius 

RI,I U, 




mm 
mm 
10" 6 

mm 

mm 
cm 


cm 


ttinc 


|90.0 


|0.5 






v disk 


|30000.0 


|1.0 






Cscat 


|0.50 


jo 10 






tente 


, |2580.00 


|1 00.00 






Rdbk 


|10.0 


|30.0 






Nphot 


on* |1 00000 


|9000 






Focus 


|30.0 


|20.0 






k 
* ave 


|10.0 


|l.O 






*■ lase 


, |7800 











Figure 6-5. Simulation with Vdisk = 30.0m/s and Vcenier =2580 xlO" 6 cm" 1 



77 

laser frequency. Larger velocities will allow better resolution of the irradiating and 
Doppler shifted frequencies. Conversely, this also implies that some minimum disk 
velocity exists at which these frequencies can no longer be resolved. Figure 6-6 
illustrates both of these implications. Shown here are signal profiles across the RIID 
detection range at two different disk velocities. Except for v cen ter and Vaisk, MCS 
conditions are identical to those in simulation 3. The signal maxima are resolved in both 
cases, but a lower disk velocity limit for baseline resolution is about 2 m/s is realized. 
The shifted profile width remains constant over an extended disk velocity range. The 
peak width was studied as a function of several other factors, including //s Cat and /4b s . In 
all cases, the peak width appeared to remain constant (within 5%). In highly absorbing 
and scattering media (high /4 bs and /4 ca , ), the total number of detected photons is greatly 
reduced. In simulation 1, several hundred photons are shown to be detected. As is the 
case in simulation 4, shown in figure 6-7, fewer that 20 photons are detected when /4 cat is 
increased. 

The image in simulation 4 has become less indicative of the shape of the moving 
object. From such an image, one could not conclude that the object is a disk; only that it 
is an object with a velocity of 30.0 m/s. This type of detection without identification 
becomes more prevalent as various parameters (such as /4 cat ) are increased and, in turn, 
the number of detected photons are decreased. For example, it has been shown that a 
moving object can be detected with a scattering coefficient greater than 30 mm" 1 . This is 
comparable to the results described by Soelkner. 10 In such an example, the detected 
number of photons may be as few as one or two, which is an indication that a moving 
object is present. The Doppler shift exhibited by the detected photons, as few as there 



78 



1600-, 



1400- 



1200-1 

5 iooo- 

X 

-^ 

■S 800- 

a 

2 600- 

o 
ss 

& 400- 



200 




t V.. , = 2.22m/s 
disk 



500 



Doppler-shifted frequencies 




1000 



1500 2000 

,-6 



^center* 10 " 6 ""'') 



V.. . =30.0m/s 
disk 



2500 



3000 



Figure 6-6. Doppler shifted profiles. 



79 




a, nc |90 


v <fck J30000.0 


|iscat |3200 


^center |2580.00 


R disk |10° 


N photons |™™ 


Focus |300 


k aver |10.0 


1 | asel J780.0 



degree 

mm/sec 

mm 

10" 6 cm -1 



L 

Cabs 
b> 

Dlasei 
Nacts 

Radius 



1.0 



10.10 



30.0 



9000 



20.0 



RHP 



width 



1.0 



I 



100.00 



10* cm " 1 



Figure 6-7. Simulation with Vdisk = 30.0m/s, Venter =2580 xlO" 6 cm" 1 , and (W = 32.0 
mm" 1 



80 

might be, must be induced by a moving object. Simulation 5, shown in figure 6-8, further 
illustrates this as a moving object is detected when both /4 cat and /4t, s are increased. 

To further demonstrate the utility of this RIID model, several of the above simulations 
were repeated for mercury at the 253.7 nm transition. The velocity of the object (disk) 
was increased to 300 m/s and the spectroscopic linewidth was increased to 1 GHz. The 
behavior was nearly identical or comparable to the 760 nm study. The ratio of detected 
photons at v center = 2.3 GHz (0.0788 cm" 1 ) to the detected photons at v center = 0, or signal 
to background ratio, is 2.4 x 10" 4 . Nevertheless, this mathematical model and simulation 
shows that the moving target is clearly detected. It is important to emphasize here that 
identification and detection are unique figures of merit. 

It is necessary to note the limitations and/or restrictions of the MCS program and the 
differences between real and simulated moving object detection. The actual spectral line- 
shape of a RIID is not rectangular, but normally Gaussian or Lorentzian. Illumination of 
a moving object by laser radiation is much more complicated than described here. 
Homogenous and isotropic scattered light is not produced from the surface of the moving 
object. Actual distances between the RIID and the moving object would be much larger 
than in these simulations. The default number of scattering events (N ac ts = 9000) can be 
much larger. The distance choice here was made for simplification and shorter 
calculation times. 

In summary, we have shown that the detection and identification of moving objects in 
turbid media with a RIID is theoretically possible. All of the results thus far are 
promising for the future development of such an imaging device. The model employed 
here is limited only by the number of photons which can penetrate the turbid 



81 




%p«a. 


atert 










tttic 
v disk 
fscat 
^center 
Rdisk 
N photons 
Focus 
* aver 
^ latet 




dejjee 

mm/sec 

mm 

lO^cm" 1 

nwn 

nwn 
cm 
nm 


?a 

L 

fabs 
tv 

D laser 
Nacts 
Radkis 




nwn 

nwn 

10" 6 cm " 1 

mm 

mm 
cm 


|90 


0.5 


|30000.0 


1.0 


|12 


12 


|2580 00 


100 00 


jioo 


300 


|300000 


9000 


|300 


200 


|10.0 


1.0 


|7800 



Figure 6-8. Simulation with Vdisk = 30.0m/s, Vcemer =2580 xlO" 6 cm* 1 , u«at = Mscai ■ 12.0 
mm" 1 






82 

medium. It has been demonstrated that a moving object can be detected in a medium 
where the scattering coefficient is greater than 35 mm" 1 . 

The model of moving object detection in turbid media described in this chapter could 
be demonstrated in several types of experiments. A particular combustion product, with 
velocity related to its size, could be imaged in the presence of various other scattering and 
absorbing particles [69]. This could provide some insight into particle formation in 
various types of flame or engines. Another application, represented by the preceding 
simulations, could be realized by the military for imaging projectiles in the atmosphere. 
A bullet or missile in the atmosphere could be imaging with overcast conditions, 
providing both velocity and structural information. 



CHAPTER 7 
FINAL COMMENTS 

Conclusions 

The mercury resonance ionization imaging detector has previously been demonstrated 
as sensitive and spectrally selective photon detector. In this work, a compact and self- 
contained Hg RIID was effectively demonstrated. This RIID was shown to be capable of 
detected fewer than 10 3 photons at 253.7 nm in non-imaging mode. When operated in 
imaging mode, the RIID described here is capable of spatial resolutions less than 80 urn. 

As with the development of any novel technology, the RIID discussed here has several 
limitations. The primary limitation of this Hg RIID is that it's susceptible to image 
distortions. Several methods have been suggested to prevent, minimize, or even 
eliminate these distortions. Also, the capabilities of the RIID as a 1 -dimensional photon 
detector are not limited by these distortions. Non-imaging mode experiments verify that, 
although spatially redistributed, the signal ions are detected. 

Future Work 

The priority for improvements to the mercury RIID is engineering quality control. 
The RIID described in this work was hampered from several unknown material 
specifications. These included the actual MCP amplification factor, the true thickness of 
the phosphor screen and metal films, the composition of the mercury dispenser, the exact 
pressure within the cell, and the quantum efficiency of the phosphor screen. A 
knowledge of several of these specifications might provide some insight into the 
unknown ionization component discussed in chapter 3. 



S3 



84 

As a method to eliminate surface charging effects, and hence image distortions, a 
thicker and more uniform Pd film should be coated on the input window. The result 
would be a more homogenous electric field between the input window and MCP. Also, 
the photoelectrons would be grounded instead of penetrating into or accumulating upon 
the surface of quartz window. Not only would the image distortions be prevented, but the 
spatial resolution might be improved upon. 

The spectral resolution of the detector in this study is defined as the absorption 
linewidth of the contained mercury vapor. Replacing this isotopic mixture with a 
monoisotopic atomic vapor, such as 202 Hg, would improve the spectral resolution of the 
RIID to about 1 GHz. This would aid in such applications as moving object detection. 

Work has already begun to employ the RIID for several of the previously mentioned 
applications including moving object detection and Raman imaging. To aid in the 
successfulness of these experiments, construction has begun on a tunable and ultra- 
narrowband Alexandrite laser system. In addition to the improved spectral resolution 
obtainable with this new laser, the overall size of the experimental setup will be reduced 
by 4-5 times. 



LIST OF REFERENCES 

[I] O. Matveev, B. Smith, J. Winefordner, Applied Optics 36 (1997) 8833. 

[2] O. Matveev, B. Smith, J. Winefordner, Applied Physics Letters 72 (1998) 
1673. 

[3] O. Matveev, B. Smith, J. Winefordner, Optics Letters 23 (1998) 304. 

[4] O. Matveev, B. Smith, J. Winefordner, Optics Communications 156 (1998) 
259. 

[5] O. Matveev, Zhulnal Prikladnoy Spektroskopii 46 (1987) 359. 

[6] R. Miles, W. Lempert, Applied Physics B 5 1 ( 1 990) 1 . 

[7] M. Smith, G. Northam, J. Drummond, AIAA 34 (1996) 434. 

[8] N. Finkelstein, W. Lempert, R. Miles, Optics Letters 22 (1997) 537. 

[9] D. Skoog, J. Leary, Principles of Instrumental Analysis, 4 th ed, Saunders 
College Publishing, Philadelphia, 1992. 

[10] D. Pappas, O. Matveev, B. Smith, M. Shepard, A. Podshivalov, J. 
Winefordner, Applied Optics 39 (2000) 491 1. 

[II] V. Letokhov, Laser Photoionization Spectroscopy, Academic Press, New 
York, 1987. 

[12] G. Magerl, B. Oehry, W. Ehrlich-Schupita, Institut fur Nachrichtentechnik 
und Hochfrequenztechnik, Conference Proceedings, Vienna, 1991. 

[13] R. Ambartzumian, V. Kalinin, V. Letokhov, Pis'ma Zhurnalu 
Eksperimental'noy i Teoreticheskoy Fiziki 13 (1971) 305. 

[14] G. Hurst, M. Payne, Principles and Applications of Resonance Ionization 
Spectroscopy, Adam Hilger Pubishers, Bristol, TN 1988. 



85 



86 



[15] G. Hurst, M. Nayfeh, J. Young, Applied Physics Letters 30 (1977) 229. 

[16] G. Bekov, V. Letokhov, O. Matveev, V. Mishin, Pis'ma Zhurnalu 
Eksperimentarnoy i Teoreticheskoy Fiziki 28 (1978) 308. 

[17] U. Brinkman, W. Hartwig, H. Telle, H. Walther, Applied Physics 5 (1974) 
109. 

[18] D. Skoog, F. Holler, T. Nieman, Principles of Instrumental Analysis, 5 th ed, 
Saunders College Publishing, Philadelphia, 1998. 

[19] V. Antonov, I. Knyazev, V. Letokhov, V. Matiuk, V. Movshev, Pis'ma 
Zhurnalu Eksperimentarnoy i Teoreticheskoy Fiziki 3 (1977) 1287. 

[20] L. Zandee, R. Bernstein, Journal of Chemical Physics 71 (1979) 1359. 

[21] S. Rockwood, J. Reilly, K. Hohla, K. Kompa, Optics Communications 28 
(1979) 175. 

[22] N. Omenetto, Journal of Analytical Atomic Spectrometry 13 (1998) 385. 

[23] O. Matveev, B. Smith, N. Omenetto, J. Winefordner, Spectrochimica Acta 
B 51 (1996) 563. 

[24] M. Shepard and J. Winefordner, Microscopy and Analysis (2000) 19. 

[25] O. Matveev, B. Smith, N. Omenetto, J. Winefordner, Applied Spectroscopy 
53(1999) 1341. 

[26] SAES Getters Incorporated, Getter-type Mercury Dispenser Products, 
http://www.saesgetters.com/prdfr_mdis.htm, December 2001. 

[27] O. Matveev, N. Omenetto, Resonance Ionization Symposium Conference 
Proceedings, American Institute of Physics, 1994, 367. 

[28] A. Podshivalov, W. Clevenger, O. Matveev, B. Smith, J. Winefordner, 
Applied Spectroscopy 54 (2000) 175. 

[29] O. Matveev, N. Zorov, Y. Kuzyakov, Journal of Analytical Chemistry 34 
(1979) 846. 






87 



[30] O. Matveev, W. Clevenger, L. Mordoh, B. Smith, J. Winefordner, in 
Resonance Ionization Symposium Conference Proceedings, American 
Institute of Physics, 1996, 171. 

[31] D. Pappas, O. Matveev, B. Smith, M. Shepard, A. Podshivalov, J. 
Winefordner, Applied Optics 39 (2000) 491 1. 

[32] Hamamatsu Corporation, Hamamatsu Product Catalog, New York (1996). 

[33] Princeton Instruments, Princeton Instruments Catalog, Trenton, NJ (1996). 

[34] U. Ellenberger, A. Glinz, J. Balmer, Measurement Science and Technology 
4(1993) 1430. 

[35] J. Ingle, S. Crouch, Spectrochemical Analysis, Prentice Hall, Upper Saddle 
River, New Jersey, 1988. 

[36] M. Grossman, R. Lagushenko, J. Maya, Physical Review A 34 (1986) 
4094. 

[37] N. Omenetto, Analytical Laser Spectroscopy, John Wiley & Sons, New 
York, 1979. 

[38] The Photonics Design and Applications Handbook, Laurin Publishing 
Company, Pittsfield, MA, 1999. 

[39] A. Podshivalov, M. Shepard, O. Matveev, B. Smith, J. Winefordner, 
Journal of Applied Physics 86 (1999) 5337. 

[40] S. Kramer, J. Young, G. Hurst, M. Payne, Optics Communications 30 
(1979) 47. 

[41] N. Karnov, B. Krynetzkii, O. Stel'makh, Kvantovaya Elecktron 4 (1977) 
2275. 

[42] R. Ambartzumian, G. Bekov, V. Letokhov, V. Mishin, Pis'ma Zhurnalu 
Eksperimental'noy i Teoreticheskoy Fiziki 31 (1974) 595. 

[43] T. Ducas, M. Littaman, R. Freeman, D. Kleppner, Physical Review Letters 
35(1975)366. 

[44] A. Smith, J. Goldsmith, D. Nitz, S. Smith, Physical Review A 22 (1980) 

577. 



88 



[45] D. Beeman, T. Calcott, S. Kramer, F. Arakawa, G. Hurst, E. Nussbaum, 
International Journal of Mass Spectrometry and Ion Physics 34 (1980) 89. 

[46] Y. Kudriavtzev, V. Letokhov, V. Petrunin, Pis'ma Zhurnalu 
Eksperimental'noy i Teoreticheskoy Fiziki 42 (1985) 23. 

[47] U. Brinkman, W. Hartwig, H. Telle, H. Walther, Applied Physics 5 (1974) 
109. 

[48] D. Bradley, C. Dudan, P. Ewart, A. Purdie, Physical Review A 13 (1976) 
1416. 

[49] R. Ambartzumian, V. Letokhov, Applied Optics 1 1 (1972) 354. 

[50] R. Ambartzumian, V. Letokhov, E. Ryabov, N. Chekalin, Pis'ma Zhurnalu 
Eksperimental'noy i Teoreticheskoy Fiziki 20 (1974) 597. 

[51] R. Ambartzumian, V. Kalinin, V. Letokhov, Pis'ma Zhurnalu 
Eksperimental'noy i Teoreticheskoy Fiziki 13 (1971) 305. 

[52] G. Hurst, M. Nayfeh, J. Young, Physical Review A 15 (1977). 

[53] M. Zimmerman, T. Ducas, M. Littaman, D. Kleppner, Journal of Physics B 
11 (1978) Lll. 

[54] W. Cooke, T. Gallagher, Physical Review Letters 41 (1978) 1648. 

[55] M. Aymar, R. Champen, C. Deslart, J. Keller, Journal of Physics B 14 
(1981)4489. 

[56] A. Mizolek, Journal of Analytical Chemistry 53 (1981) 118. 

[57] P. Dyer, G. Baldwin, C. Kittrel, D. Imre, E. Abramson, Applied Physics 
Letters 42 (1983) 311. 

[58] D. Andrews, Lasers in Chemistry, Springer-Verlag, New York, 1997. 

[59] Oriel Instruments, The Book of Photon Tools, Oriel Instruments, Stratford, 
CT, 2001. 

[60] Handbook of Optics, Optical Society of America, McGraw-Hill, New 
York, 1978. 



89 



[61] E. Saloman, Spectrochimica Acta B 43 (1991) 319. 

[62] M. Shepard, J. Temirov, O. Matveev, B. Smith, J. Winefordner, Optics 
Communications, in press (2002). 

[63] A. Podshivalov, W. Clevenger, 0. Matveev, B. Smith, J. Winefordner, 
Applied Spectroscopy 54 (2000) 175. 

[64] P. Potts, A Handbook of Silicate Rock Analysis, Chapman & Hall, New 
York, 1987. 

[65] K. Kanaya, S. Okayama, Journal of Physics D 5 (1972) 43. 

[66] A. Jelalian, Laser Radar Systems, Artech House, Boston, 1992. 

[67] N. Parikh, J. Parikh, Optics and Laser Technology 34 (2002) 1 77. 

[68] K. Ogawa, IEEE Transactions on Nuclear Science 44 ( 1 997) 1521. 

[69] T. Histen, O. Guell, I. Chavez, J. Holcombe, Spectrochimica Acta B 51 
(1996) 1279. 












BIOGRAPHICAL SKETCH 
Michael Shepard was born on September 23 rd , 1974, in Asheville, North Carolina. He 
is the son of Deena Duncan and Charles Shepard. He lived most of his childhood in the 
mountains of western North Carolina and graduated from McDowell County High School 
in 1993. Michael then graduated from Western Carolina University in 1997 with a 
Bachelor of Science degree in chemistry. Michael attended graduate school at the 
University of Florida and completed his doctoral research in 2002 under the direction of 
Dr. James D. Winefordner. 



90 



I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and quality, 
as a dissertation for the degree of Doctor of Philosophy 



Jimes D. Winefordner/Chairman 
Professor of Chemistry 



I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and quality, 
as a dissertation for the degree of Doctor of Philosophy. 




Samuel 0. Colgate 
Professor of Chemistry 



I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and quality, 
as a dissertation for the degree of Doctor of Philosophy. 



t l ^>' 



Anna Brajter-Toth 

Associate Professor of Chemistry 



I certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and quality, 
as a dissertation for the degree of Doctor of Philosophy 




Weihong Tan 

Associate Professor of Chemistry 



1 certify that I have read this study and that in my opinion it conforms to 
acceptable standards of scholarly presentation and is fully adequate, in scope and quality, 
as a dissertation for the degree of Doctor of Philosophy 




F Eugene Dunnam 

Professor of Physics and Astronomy 



This dissertation was submitted to the Graduate Faculty of the Department of 
Chemistry in the College of Liberal Arts and Sciences and to the Graduate School and was 
accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy 



August 2002 



Dean, Graduate School 



LD 
1780 



O 



[20£ 



UNIVERSITY OF FLORIDA 

lllllllill 

3 1262 08555 3559